done??HEADmaster

1 files changed, 97 insertions, 0 deletions

diff --git a/sorsyl/lib/syllable.ml b/sorsyl/lib/syllable.ml
new file mode 100644
index 0000000..ef72736
--- /dev/null
+++ b/sorsyl/lib/syllable.ml
@@ -0,0 +1,97 @@
+(** Module for representing syllables and their components *)
+
+open Base
+
+(** Type representing a syllable with its phonological components *)
+type t = {
+  onset : string;      (** Initial consonants *)
+  medial : string;     (** Medial consonants (between onset and nucleus) *)
+  nucleus : string;    (** Vowel core of the syllable *)
+  coda : string;       (** Final consonants *)
+  tone : string;       (** Tonal information *)
+  spelling : string;   (** Orthographic representation *)
+  start_idx : int;     (** Start position in the word *)
+  end_idx : int;       (** End position in the word *)
+  stressed : bool;     (** Whether this syllable is stressed *)
+}
+
+(** Create an empty syllable *)
+let empty = {
+  onset = "";
+  medial = "";
+  nucleus = "";
+  coda = "";
+  tone = "";
+  spelling = "";
+  start_idx = 0;
+  end_idx = 0;
+  stressed = false;
+}
+
+(** Create a syllable with specified components *)
+let create ?(onset = "") ?(medial = "") ?(nucleus = "") ?(coda = "") ?(tone = "") 
+           ?(spelling = "") ?(start_idx = 0) ?(end_idx = 0) ?(stressed = false) () =
+  { onset; medial; nucleus; coda; tone; spelling; start_idx; end_idx; stressed }
+
+(** Get the complete syllable string *)
+let all t = t.onset ^ t.medial ^ t.nucleus ^ t.coda ^ t.tone
+
+(** Get the rhyme (medial + nucleus + coda) *)
+let rhyme t = t.medial ^ t.nucleus ^ t.coda
+
+(** Check if the nucleus contains length marker *)
+let is_long t = String.is_substring t.nucleus ~substring:"ː"
+
+(** Finalize a syllable - determine if it's stressed based on position relative to stress marker *)
+let finalize t ~end_idx ~stress_idx =
+  let stressed = 
+    if stress_idx >= 0 then
+      stress_idx >= t.start_idx && stress_idx <= end_idx
+    else
+      false
+  in
+  { t with end_idx; stressed }
+
+(** Pretty print a syllable with Unicode box drawing *)
+let pretty_print t =
+  let open Printf in
+  let box_width = 10 in
+  let pad s = 
+    let len = String.length s in
+    if len >= box_width then String.prefix s box_width
+    else s ^ String.make (box_width - len) ' '
+  in
+  
+  let h_line = String.make box_width '-' in
+  let top = "+" ^ h_line ^ "+" in
+  let bottom = "+" ^ h_line ^ "+" in
+  let make_row label content = sprintf "|%s|" (pad (label ^ content)) in
+  
+  String.concat ~sep:"\n" [
+    top;
+    make_row "σ: " (all t);
+    make_row "O: " t.onset;
+    make_row "M: " t.medial;
+    make_row "N: " t.nucleus;
+    make_row "C: " t.coda;
+    make_row "T: " t.tone;
+    if t.stressed then make_row "* " "stressed" else "";
+    bottom
+  ] |> String.strip
+
+(** Convert to string representation *)
+let to_string t =
+  Printf.sprintf "{onset=%S; medial=%S; nucleus=%S; coda=%S; tone=%S; stressed=%b}"
+    t.onset t.medial t.nucleus t.coda t.tone t.stressed
+
+(** Append to onset *)
+let append_onset t s = { t with onset = t.onset ^ s }
+
+(** Append to nucleus *)
+let append_nucleus t s = { t with nucleus = t.nucleus ^ s }
+
+(** Append to coda *)
+let append_coda t s = { t with coda = t.coda ^ s }
+
+(** Append to tone *)
+let append_tone t s = { t with tone = t.tone ^ s }
\ No newline at end of file